weights. The coefficients of the EVI equation for Landsat Enhanced Thematic
Mapper (ETM+) and MODIS bands are L = 1; C 1 = 6 and C 2 = 7.5 (Huete et al.
2002). A greater weighing of the NIR band extends greenness sensitivity in high
biomass areas, while the Blue reflectance term stabilizes aerosol influences and
mis-corrections of Red reflectance retrievals, as originally developed in ARVI.
The NIR and Red reflectances for a constant canopy (e.g. any fixed LAI and
structural arrangement) with variable underlying backgrounds is represented with
a line connecting the various background-dependent spectra, with a slope and Yintercept specific to that value of LAI (Baret and Guyot 1991; Huete 1988). This
line of constant vegetation is known as a ‘biophysical isoline’, and includes the
information needed to minimize or remove the background influences from vegetation indices. We can distinguish between biophysical isolines and VI isolines
with a SAIL radiative transfer model simulation of NDVI isolines (Fig. 1.3b). The
discrepancies between the two types of isolines are related to optical influences
other than the vegetation layer, and provide methods for improving or designing
optimal VIs so as to align the VI isolines with the biophysical isolines.
Canopy background signals are inherent to most canopies, and are non-linearly
coupled to the vegetation signal to various extents dependent on the structural
arrangement of the vegetation in a canopy. Red and NIR transmittance (extinction)
through a photosynthetically-active canopy differs significantly with much higher
optical thickness in the Red, due to the highly absorptive properties of leaf pigments, and relatively low NIR optical extinction due to the highly scattered
(transmitted and reflected) signal (Fig. 1.4).
A primary difference among ratio-based, linear combination, and optimized
indices is in the ‘L’ parameter, which optimizes the measure of greenness through
a simple, first-order application of Beer-Lambert’s law to describe differential red
and NIR extinction through vegetation canopies,
q c ¼ q v þ t
2
c q s
ð1:9Þ
where canopy reflectance, q c , is the sum of the vegetation layer reflectance, q v , and
the two-way canopy transmitted- soil reflected signal, t
2
c q s . When ‘L’ is optimized
correctly, the VI signal becomes blind to soil background variations as the relative
optical depths of the two bands are more closely adjusted to only see the greenness
of the canopy. The NDVI represents the special case of L = 0, in contrast to
L = 1 in the EVI. Larger values of L (L [ 10), on the other hand, approximate the
linear combination and SMA derived indices.
The use of multiple VIs offers a more complete characterization of canopy
properties. The PVI, EVI and TC greenness are more spectrally sensitive to the
NIR and will contain information from multiple canopy leaf layers, due to the
higher canopy optical penetration depths. In contrast, the NDVI is most sensitive
to the canopy-absorbing Red band and will sense primarily the uppermost leaf
layers. As a result, the first set of indices presents higher canopy penetration,
allowing extended sensitivity over higher LAI/biomass areas where the NDVI
saturates (Fensholt et al. 2004; Huete et al. 2006).
8
A. Huete et al.
Mapper (ETM+) and MODIS bands are L = 1; C 1 = 6 and C 2 = 7.5 (Huete et al.
2002). A greater weighing of the NIR band extends greenness sensitivity in high
biomass areas, while the Blue reflectance term stabilizes aerosol influences and
mis-corrections of Red reflectance retrievals, as originally developed in ARVI.
The NIR and Red reflectances for a constant canopy (e.g. any fixed LAI and
structural arrangement) with variable underlying backgrounds is represented with
a line connecting the various background-dependent spectra, with a slope and Yintercept specific to that value of LAI (Baret and Guyot 1991; Huete 1988). This
line of constant vegetation is known as a ‘biophysical isoline’, and includes the
information needed to minimize or remove the background influences from vegetation indices. We can distinguish between biophysical isolines and VI isolines
with a SAIL radiative transfer model simulation of NDVI isolines (Fig. 1.3b). The
discrepancies between the two types of isolines are related to optical influences
other than the vegetation layer, and provide methods for improving or designing
optimal VIs so as to align the VI isolines with the biophysical isolines.
Canopy background signals are inherent to most canopies, and are non-linearly
coupled to the vegetation signal to various extents dependent on the structural
arrangement of the vegetation in a canopy. Red and NIR transmittance (extinction)
through a photosynthetically-active canopy differs significantly with much higher
optical thickness in the Red, due to the highly absorptive properties of leaf pigments, and relatively low NIR optical extinction due to the highly scattered
(transmitted and reflected) signal (Fig. 1.4).
A primary difference among ratio-based, linear combination, and optimized
indices is in the ‘L’ parameter, which optimizes the measure of greenness through
a simple, first-order application of Beer-Lambert’s law to describe differential red
and NIR extinction through vegetation canopies,
q c ¼ q v þ t
2
c q s
ð1:9Þ
where canopy reflectance, q c , is the sum of the vegetation layer reflectance, q v , and
the two-way canopy transmitted- soil reflected signal, t
2
c q s . When ‘L’ is optimized
correctly, the VI signal becomes blind to soil background variations as the relative
optical depths of the two bands are more closely adjusted to only see the greenness
of the canopy. The NDVI represents the special case of L = 0, in contrast to
L = 1 in the EVI. Larger values of L (L [ 10), on the other hand, approximate the
linear combination and SMA derived indices.
The use of multiple VIs offers a more complete characterization of canopy
properties. The PVI, EVI and TC greenness are more spectrally sensitive to the
NIR and will contain information from multiple canopy leaf layers, due to the
higher canopy optical penetration depths. In contrast, the NDVI is most sensitive
to the canopy-absorbing Red band and will sense primarily the uppermost leaf
layers. As a result, the first set of indices presents higher canopy penetration,
allowing extended sensitivity over higher LAI/biomass areas where the NDVI
saturates (Fensholt et al. 2004; Huete et al. 2006).
8
A. Huete et al.